Purpose: To determine the appropriate patients, methods, and timing for screening for diabetic retinopathy.
Data sources: Relevant articles were identified through prominent review articles, the authors' files, recommendations from experts, and a MEDLINE search (1986 to the present); additional references were selected from the bibliographies of identified articles.
Study selection: Selection of articles on the natural history of retinopathy was limited to large clinical series and formal epidemiologic studies of defined populations. Selection of articles on the therapeutic effect of photocoagulation and of glycemic control was limited to randomized trials. Sources bearing on the accuracy of screening modalities were necessarily more varied.
Data extraction: For important variables, individual estimates from multiple studies are presented rather than a single meta-analytic summary estimate.
Results: Screening for retinopathy is justifiable if early detection leads to less vision loss at an acceptable cost. The evidence shows that 1) laser therapy reduces the rate of vision loss by 50% among patients with proliferative retinopathy and macular edema, conditions that are often asymptomatic; 2) duration of diabetes is the main risk factor for retinopathy; and 3) standard ophthalmoscopic examination has only moderate sensitivity (about 80% in research settings) and specificity (greater than 90% for proliferative retinopathy but lower for macular edema), making seven-field stereophotography a more accurate method. Estimates of cost effectiveness indicate that screening for retinopathy not only saves years of vision but may be cost saving from a societal perspective.
Conclusions: Screening for retinopathy in patients with diabetes, and subsequent photocoagulation therapy for those who have high risk macular edema or proliferative retinopathy, is clearly beneficial.